The gold standard for making therapeutic judgments is the randomized controlled trial. However, a single trial by itself is rarely definitive even if it is properly designed and has adequate sample size. The practitioner is therefore faced with the need to amalgamate the results of a number of similar trials and to make an informed judgment concerning the value of a therapeutic approach. The usual approach is via meta-analysis in which results from a number of trials are pooled. A major limitation in this approach has to do with dependence on the published literature, where selective factors may influence publication strategies and decisions. Thus use of meta-analysis may lead to biased judgement, if the "file drawer" problem (unpublished results) is large or the results in the file drawers are different from those published. In order to address this problem we plan: 1. An assessment of published reports of perinatal clinical trials to gauge their robustness against bias in data collection and reporting processes, 2. A survey of departments of perinatal medicine to assess the magnitude of the file drawer problem, 3. A comparison of results of published and unpublished trials as a means of assessing the nature of the bias represented by the file drawer problem. In addition, an ongoing inventory of clinical trials in the perinatal field will be established to provide a continuing database for evaluation of operating features of trials in the field. The aim will be to use it as a resource for the design and conduct of future trials and for gauging the magnitude of the "file drawer" problem in the future.